Automatic Patient Matching for Genome-Guided Adoptive Cell Immunotherapy Trials at a Large Health System

Document Type


Publication Date


Publication Title



oregon; portland; chiles; genomics


Introduction: As precision cancer therapies and genome-guided immunotherapies are trending to increasing biomolecular specificity, rapid and routine patient screening is essential to quickly identify the subset of patients that meet highly specific enrollment criteria. Here we developed a protocol for rapid clinical genomic data reanalysis, and patient identification for supporting precision medicine trials across a large healthcare system. We applied these approaches to extract human-leukocyte antigen class I (HLA-I) and tumor mutation information from next-generation sequencing data to support a novel investigator-initiated trial that uses the adoptive transfer of T-cells engineered to express T-cell receptors (TCRs) that target hotspot mutations. Methods: Electronic health record and genomic data from patients (n=8,115) spanning 33 cancer subtypes and with at least one reported pathogenic mutation were mined to identify existing patients that might qualify for an ongoing adoptive cell therapy matching clinical trial. Patients received comprehensive genomic profiling (CGP) using the Illumina TruSight Oncology 500 High Throughput assay (TSO500 HT) within the Providence Health system between September 2019 and May 2022. Previously reported pathogenic mutations were compiled and BAM files from patients with a pathogenic KRAS G12V/D mutation and reportedly not deceased were HLA-typed using DRAGEN HLA Caller. Key criteria to qualify for the clinical trial includes the presence of a pathogenic KRAS G12D mutation with either HLA-A*11:01 or HLA-C*08:02 match, or a pathogenic KRAS G12V mutation with HLA-A*11:01 match, while also assessing feasibility via last known contact date with the patient and other variable factors. Results: From the 8,115-patient cohort who received CGP testing, 960 patients (11.8%) harbored a pathogenic KRAS G12V/D mutation. The most enriched tumor types include pancreas (28.5%), bowel (25.4%), lung (18.8%), cancer of unknown primary (9.7%), and bladder/urinary tract (4.2%). As of early May 2022, 58.1% (n=558) of patients with a pathogenic KRAS G12V/D were not reported as deceased, and reanalysis for HLA typing was performed on CGP datasets. We identified 17.4% (n=97) of alive KRAS G12V/D patients had an HLA-A*11:01 or HLA-C*08:02 genotype matching the trial eligibility. Conclusions: Although KRAS G12V/D is a relatively common solid tumor driver mutation, adding an HLA matching criterion significantly restricts the number of eligible patients. As more and more neoantigen-targeting TCRs are identified, automated matching algorithms represent a necessary step in rapidly identifying patients who might benefit from highly targeted immunotherapeutics.

Clinical Institute



Earle A. Chiles Research Institute